Merge #2213

2213: Hir generic param r=flodiebold a=matklad

r? @flodiebold 

This should make the life of IDE easier: before, it got `GenericParam(u32)` which was of questionable utility. Now, it's a proper code_model type, so it can gain `source`, `name`, `module` and all the other hir methods, should the IDE need them. Moreover, IDE now doesn't care about internal representation of generic param, which seems like a long-term win. 

The problem is, of course, that we now have to types named `GenericParam` in hir: this code_model type, and an internal type with an index which doesn't know about the parent. I think it's fine for the time being, but, after we finish cratefication of hir, this local `GenericParam` should move to `hir_def` or `hir_ty`, and *maybe* restrucured as `ParamId / PramData` pair. 

Co-authored-by: Aleksey Kladov <[email protected]>

1 files changed, 7 insertions, 1 deletions

diff --git a/crates/ra_hir/src/code_model.rs b/crates/ra_hir/src/code_model.rs
index 09c4e97fa..5a0bd0c19 100644
--- a/crates/ra_hir/src/code_model.rs
+++ b/crates/ra_hir/src/code_model.rs
@@ -23,7 +23,7 @@ use crate::{
     adt::VariantDef,
     db::{AstDatabase, DefDatabase, HirDatabase},
     expr::{validation::ExprValidator, BindingAnnotation, Body, BodySourceMap, Pat, PatId},
-    generics::HasGenericParams,
+    generics::{GenericDef, HasGenericParams},
     ids::{
         AstItemDef, ConstId, EnumId, FunctionId, MacroDefId, StaticId, StructId, TraitId,
         TypeAliasId,
@@ -1121,3 +1121,9 @@ impl Local {
         src.map(|ast| ast.map(|it| it.cast().unwrap().to_node(&root), |it| it.to_node(&root)))
     }
 }
+
+#[derive(Clone, Copy, Debug, PartialEq, Eq, Hash)]
+pub struct GenericParam {
+    pub(crate) parent: GenericDef,
+    pub(crate) idx: u32,
+}